Spot on! This is kinda what I was trying to say in my original rather disappointed post way back up near the top of this thread but you have explained it far better than I did! Especially now I know it doesn’t take workout alternates into account.
I regularly increase the duration of my prescribed Sunday endurance ride. Not by a ridiculousl amount but by an amount I know I can handle just fine, especially outdoors. If RLGL isn’t considering workout alternates (up or down in duration or Pl) then I’d expect that to be fixed before it goes full release.
If I had followed RLGL days since my plan started in December I reckon my fitness would be lower than it is today (by all measures, not just CTL) and I know I’m not over trained. I appreciate it is likely erring on the side of caution so I’ll continue to keep an eye on it while also monitoring all the other factors that TR can’t see… As others have said above, we shouldn’t rely solely on the robots to tell us what to do!
That’s just a user interface issue. The point of this technology is to provide some guidance to the user and help dial in the exact right load each day.
Here’s what I think would make sense:
Adaptations are never suggested more than a week out (I’m not sure the current timeframe)
You always see the original plan, any modifications you made and the suggested adaptation
Adaptations are able to be accepted or rejected for each workout independently and even if you don’t accept right away you still can later
When you go to make an alteration you can specify that you want to alter the duration or PL and why (post-ride survey style, except while choosing the workout alternate)
The interface should indicate why an alternate was suggested (weekly TSS too high, poor sleep, HRV too low, etc)
Wouldn’t that be way better? You as the user would feel like you understand what’s happening and why. You can provide the algorithms all the data they need to understand the difference between reductions in training due to fatigue vs other factors and better tune the adaptations. They could even start making some more sophisticated suggestions for training plan volume and other factors. The cleaner and more detailed the data in, the easier it is for the model to give good output. Right now the reasons for users adapting their plans manually is completely obfuscated to the model. That’s bad.
Sure, but it will NEVER happen. Your expectations are very high and you will be disappointed. All you’ll likely ever get is the ability to choose RLGL’s level of moderation. According to Nate, in the future you will able to choose wether you want RLGL to be conservative, moderate or liberal. Maybe, just maybe, TR will also separate RLGL adaptions and ML adaptions, but that’s about it for the reasonable future.
I think we need to select alternates the day of, or the day before, after your last workout. That way RLGL would be incorporated into the alternate selection model.
The critical part is really just always showing the adaptations and letting the user accept or reject the adaptations individually. They already have all the UI work done to show adaptations. They just need to let you accept or reject individually. It’s fairly trivial and doesn’t require changing the algorithms.
Adding the survey upon manual modification is pretty low hanging fruit too. Using that data is way more complicated, but they can take their time on that. Getting the data available to them should be done ASAP, IMO.
The more resources we expect TR to put into making RLGL our version of what it should/could be, the longer it will take to bring out WLV2 (unstructured rides). I don’t see TR making that trade. Better to bring an additional feature to market.
Personally, I’d like to be able to accept RLGL adaptations separately. But like most things, I can do my own work around.
I am using Intervals.icu, athletica and elevate to compare vs RLGL. It is curious that while at elevate and intervals.icu I am in an optimal zone, I was at RED with RLGL on Sunday. I did a 3hrs z2 ride that day and I could feel my legs a little weak, specially on those sudden accelerations from the group when resuming after a red light. I’m not an expert so I am not sure if that is the feeling I should feel on an optimal zone or it was an indication that I should rest instead as indicated by the RLGL.
Saturday and Sunday I had two hard rides and understandably, RLGR gave me a red day Monday (today). I was planning on taking today off anyways, so that’s fine. However, I also got a yellow day tomorrow, and I had a VO2 max workout scheduled. My options as I see them are:
Trade the VO2 for endurance tomorrow and proceed with the rest of my training plan for the week (No VO2 workout for the week)
Swap VO2 for endurance and do the VO2 workout Wednesday.
Take a second rest day in a row Tuesday and do my VO2 workout Wednesday.
Ignore the yellow light and just do the VO2 workout Tuesday.
I’m tempted to do #2, as that keeps my volume up, but that means I’m doing my VO2 workout after an endurance day instead of a rest day. Pushing my VO2 workout further back in the week would disrupt my plan too much, so I don’t see that as an option.
This is a useful feature for me, although too be honest it is mostly confirmation of what I know/feel already.
I think it could be much improved if there is a way to include Garmin’s “body battery” status into the calculation, my experience is that body battery (esp. recovery rate at night) is very good indication of fatigue or (impeding) illness.
Lastly, it’s a shame TR can’t distinguish between my “regular” bike rides and the ones I do in a velomobile; a 200km ride in my VM is about 2/3 the TSS of a regular 200km and recovery is much faster; I would rate the day after as “yellow” instead of “red”.
Do you have a power meter on the velomobile or is it HR-estimated TSS?
Why would it be physically less stress to output the same power and duration just because you have better CdA? Presumably you’re using almost identical leg muscles removed vs upright. Your vote gets spared, but I doubt that factors into aerobic and leg muscle stress. I guess it could affect RPE, but that’s probably misleading in this case.
First, if your FTP and TSS for every workout in all systems is not identical in the past and present, you can’t compare. Also, intervals.icu is using CTL/ATL/TSB and TR is not. You’re in the Red with intervals.icu when your Form is -30 and below. TR will never allow you to dig that deep of a hole and would warn you way before that with Red days. RLGL appears at the moment to be conservative erroring on the side of caution.
EDIT: Additionally, intervals.icu doesn’t take into account massive days regarding being in the “Optimal” training zone, it’s only looking at Form. So you could have a big 8 hour 400 TSS day and still be in the optimal zone for the next day. TR on the other hand would mark that next day Red as it should. Form is not the whole story and shouldn’t solely be relied upon for next day’s training.
I have recently been following a polarised build plan, and today I had the green light to train. So with bearing in mind I had a vo2 max session planned tomorrow I decided to do a very short and steady Z1/Z2 ride with a coffee stop in the middle. Tomorrow is now showing as a yellow day and has suggested that I swap the vo2 session to an endurance session. Is this correct??
I’ll have to go back and enter/correct all my runs for hrTSS. Cant say I’ll go back further than January though.
—-
Edit - Done
I’m only running once a week on average so easy to copy hrTSS from TrainingPeaks but it doesn’t look as though it’s made any difference to recent RLGL flags.
IME thats how its operates and its a little too dominated by TSS. For instance for me on Saturday I went for a cafe ride the second half was a bit more intense so it made Sunday Amber but Monday was Green (White). I went for a longer walk on Sunday and that was enough to make the Monday become Amber too. Ive not had it adapt down yet but I believe that would be the intention too.
Whether you have the workout pre-planned or not will give you the same result with RLGL – instead, thinking about it like you seemed to be doing in your initial comment is the best way to approach it, IMO.
Consider what your workout the day after will be like, and use that to decide where to add volume to your plan. Often, on days where you have a rest/recovery day the day after will be your best bet.
Seems like we had a mixup of information here – if you have runs being imported, RLGL will take them into account whether or not you have TSS entered. No need to manually enter in TSS for past runs – sorry about that!!
You are right, it is not an apples to apples comparison, but at the end I just want to understand when I can still push it to get improvements instead of wasting the day resting. Intervals uses their own algorithm made FTP calculation based on the rides information, no input from my side to enter the number. Elevate number is a manually input number which is not that far from the last AI calculated from TR. And Garmin is not using, I guess, the ftp to indicate if it was a productive green day or not. Since I am stagnating at the similar FTP for a year I am worrying if I am giving a good effort or not, and normally elevate had been in the past a good indicator. And my worries are how much confidence I must give to the RLGL red, because it seem a little, tiny bit conservative. Is that good or bad? maybe in some cases red is good, but in others may reduce your progress.
Yes and no. I definitely had a big change in my training, as you and I have both stated. But I also was very careful to keep TSB almost always optimal. as well as CTL ramp. So it aligns with broadly accepted training advice. I know that RLGL has more factors, but this is training at a level that I’ve successfully done before and my TR account has my history of that. It seems RLGL might be too conservative in this scenario.
Regarding the outdoor ride matched to Gibraltar -1, you’re right that the ride was all over the place regarding zones. Thus, there wasn’t a good match in the TR library that I could find. However, I rode for more than 90 minutes at or above tempo on that ride. Thus, I clearly am capable of 90 min tempo. So my thinking was to update my PL to reflect what I can do.
I’m not saying it isn’t useful or doesn’t work for some common scenarios. I do question though what it’s supposed to be indicating. Is it supposed to be an indicator of overtraining or overreaching, per the definitions below (from Training and Racing with a Power Meter, 3rd ed):
Overreached An acute state of fatigue and hence diminished performance resulting from a brief period of excessive training relative to what you normally perform. Although many times riders describe themselves as being “overtrained,” in reality they have usually simply overreached and their performance will recover after just a few days of rest or reduced training.
Overtrained A chronic state of overreaching from which recovery takes a long period of time, usually over 30 days. Also called Overtraining Syndrome (OTS).
TR seems to be pitching RLGL as doing both?
But then here it seems to focused on long term:
But looking at the implementation it seems that it’s basically a short-term (overreach) indicator, not a long-term (overtraining) indicator. I know these terms get misused by others, so it’s not unexpected that TR would too, but the message is that it’s both?
If RLGL is supposed to cover both scenarios, then I think the messaging needs to be clear and the terminology cleaned up and the UI needs to distinguish between a yellow or red day due to a recent hard day or too much training stress overall.
I’m hoping for clarification on that specifically.
